Data collected from the monitoring system installed in the St. Anthony Falls Bridge, a posttensioned concrete box girder bridge in Minneapolis, Minnesota, represented a combination of temperature-dependent effects due to daily and seasonal temperature changes and time-dependent effects related to creep and shrinkage of the concrete. In the companion paper of this study, the time-dependent creep and shrinkage effects were isolated from the data, and the rates of the creep and shrinkage were adjusted according to the Arrhenius equation. In this paper, results from finite-element models incorporating a variety of time-dependent provisions were compared with the measured time-dependent behavior. The modeling methodology is discussed extensively, including the procedures for incorporating the segmental construction staging sequence and midspan closure pour. Each of the models provided different predictions, and none were found to provide reliable estimates of the measured results throughout the entire 5-year period of investigation, although this was not unexpected due to the uncertainty inherent in the long-term predictions of the time-dependent behavior of concrete structures. Provisions from several standards with asymptotic creep models underestimated the longitudinal deformations and were found to approach asymptotic creep and shrinkage values before any indication of asymptotic behavior was observed in the measured data, although early-age behavior was accurately captured by these provisions. The shape of available logarithmic creep models appeared to be consistent with the measured data, although these models overestimated the longitudinal deformations of the bridge. The best predictions of the measured data were given by the provisions of a previous model code with an asymptotic creep model that properly accounted for the large volume-to-surface ratio of the bridge. Investigation of the computed vertical deflections showed that for all models the direction of the deflections reversed after completion of the structure, and that this bridge was not in danger of failure from excessive deflections. This was believed to be due to how continuity of the structure was achieved and the balance between the posttensioning forces and the gravity loading. Finally, Bayesian regression was proposed to update the finite-element model predictions to better match measured data. This technique could be used in structural monitoring applications to detect when time-dependent deformations fall outside expected bounds.
|Original language||English (US)|
|Journal||Journal of Bridge Engineering|
|State||Published - Jul 1 2017|
- Finite-element modeling
- Posttensioned concrete bridges